Note: Norm was the principle scribe on Thursday, Jim on Friday. The chat logs appear to have been incompletely captured. These minutes were constructed partly by hand by Norm.
Henry:We may need more time for the profiles document
Norm:We could conceivably have the whole of Friday afternoon if we moved the other stuff around a bit.
Henry:Far enough in the future that we don't have to plan it today.
Norm:Colocated with XML Prague is probably the next obvious opportunity.
... Anyway, we can focus on that later
< scribe> ACTION:A-237-01 Liam to update the charter page or link from the Processing Model WG homepage [recorded in http://www.w3.org/2013/09/26-xproc-irc]
Accept this agenda?
09 Oct 2013, accepted.
Alex raised some questions about alignment with serialization in the XQuery/XSLT 3.0 specs
Norm proposes retitling the "Abandon support for XPath 1.0" to "Align with XQuery/XSLT 3.0 specifications"
Alex:The new template stuff in XSLT 3.0, the ability to use curly braces in element content, is a lot like our p:template
... I also wonder about the issue I raised a while back:
Alex:It would be unfortunate to have different semantics than XSLT
< scribe> ACTION:A-237-02 Norm to review the curly-brace template stuff in XSLT wrt how it compares to p:template [recorded in http://www.w3.org/2013/09/26-xproc-irc]
Alex asks about step inventories
Norm:I think we should have a section where we consider new steps as part of the spec: promoting p:template and friends is one case, adding p:zip/unzip is another that occurs to me.
Some discussion of breaking the spec into two parts.
Some discussion of how versioning would would work.
Norm:Is there anyone opposed, in pricinple, to having two specs?
Jim:I'm not against it, but I wonder if it'll be better for users
Alex:What about writing a primer?
Norm:That would have to be in our deliverables if we put it on the REC track.
< jf_2013> Alex proposal for non xml docs http://lists.w3.org/Archives/Public/public-xml-processing-model-wg/2012Oct/0006.html
< jf_2013> Vojtechs proposal http://lists.w3.org/Archives/Public/public-xml-processing-model-wg/2012Sep/0020.html
Henry:We've discussed but never come to conclusion on the question of adding the XQuery/XSLT invariant to XProc
The balance of Thursday was spent discussing various notes and drafts related to the requirements for V.next. Ultimately, the WG produced XProc V2.0 Requirements which it intends to publish as soon as possible.
The following random items of discussion are recorded:
Some discussion of ‘standalone'. General agreement that it's not useful enough to warrant any further consideration.
Some discussion of adding validation. Can it be represented as a decision tree? Not really, it's a lattice.
Norm scribbled the following diagrams on the board during this discussion.
Henry goes to the whiteboard HT: I take it that you are not allowed to do xinclude if you choose the basic profile Alex: we should add sentence to section 2 Norm: dtd validation cant be done on infoset HT: I have seen ppl write docs which have dtd's which are meant to apply after xinclude HT: dont want to require ppl to rewrite dtd where xinclude is allowed and where it isnt HT: you could not do it in a pipeline, you would need resource manager Norm: I suppose you can, but its not practically useful Norm: an entity reference in pcdata would not be an entity reference HT: what this means is, we just describe the set of sentences w/o the grammar HT: this doesnt talk about ordering and it needs too scribe losing some of the flavour of the conversation, whiteboard battles ensue Henry v Norm <alexmilowski> You are in a dark room with an XML document. You see a DTD and an XML Schema. <alexmilowski> LOOK <alexmilowski> You see an XML Document, a DTD, and an XML Schema. There is a door on the west wall and a hallway going east. <alexmilowski> VALIDATE WITH DTD <alexmilowski> The DTD contained a fatal dose of parameter entities. You died.
ACTION:A-237-03 Henry to deal with “23603” and “23606” (remove old school capitalization).
HT: this is the serious one about standalone (4.1) Alex: we went down this road and said this doesnt work HT: no one trusts it Norm: all of these profiles explicitly ignore standalone, as it is widely untrusted HT: there is nothing a processor can do with standalone decl HT: pretty sure, that the value of standalone has no impact on processor behaviour and therefore it can't have impact on infoset content, ergo not relevant (HT attempts to revalidate this thinking) Alex: reminds us in the XML spec 'The standalone document declaration MUST have the value "no" if any external markup declarations contain declarations of:' HT: what he might want, taken literally ... treat standalone="no" as distinct from standalone attribute being absent HT: and thats contra the spec HT: I think thats true Alex: whats the default of standalone ? Norm: its 'If there are no external markup declarations, the standalone document declaration has no meaning. If there are external markup declarations but there is no standalone document declaration, the value "no" is assumed.' HT: we dont enforce any validity constraints in the profiles that dont involve validation
ACTION:A-237-04 Norm to draft response to 4.1 CMSCQ comments
ACTION:A-237-05 Henry to manage response to 4.3
ACTION:A-237-06 Henry to draft response to comment 4.4
HT: If CMSCQ wants a reference I will put it in Norm: he wants us to do an analysis of existing processors and match up profiles HT: we are ok to fill a small gap in the set of challenges who are writing specs for xml applications
ACTION:A-237-07 Henry to respond to CMSCQ comment 4.5
HT gets back to standalone HT: the only way an error arises, the only document that violates the standalone doc validity constraint is one that says standalone="yes" but isn't (has something external that matters) Alex: and you need a validating processor to tell this HT: what that means is that with the default you never get an error HT: its the opposite, its going to obsure errors where there are some HT: for our purposes we are done Norm: I think he is asking for is different behaviours, which is somewhat invalid HT: basic or full profile processors should produce and error i the presence of standalone="no" (Norm: only in the presence of dtd validation) HT: What Michael is suggesting is standalone="no", that a basic or id level processor should throw an error HT: you've said 'I need external information' ... thats what standalone="no" means HT: its a kind of profile error HT: the top level goal is to fix the xml spec - and we can't do that (requires a kind of profile level error)
Back to XProc and requirements and use cases
ACTION:A-237-08 Norm to editorialize usage of the term 'connection'
ACTION:A-237-09 Norm will send through 'we are dropping parameters' email
Working Group records its thanks to Henry and the University for magnificent hosting. Henry especially for the most excellent bacon rolls and cookies!http://www.w3.org/2013/09/26-xproc-irc]